(185) Trends in Documentation of Stigmatizing Language for Addiction from 2013 to Present

Background & Introduction: Stigma for those with substance use disorders can take many forms and can result in numerous negative effects, from discouraging those with substance use disorders from seeking medical care to limiting the willingness of policymakers to prioritize funding. Even among mental health providers and other healthcare workers, stigma exists based on the language used to refer to those with substance use disorders. A randomized study of a clinical vignette in two versions, one using the term “substance abuser” and the other using the term “substance use disorder,” found that when mental health providers received the vignette with stigmatizing language, it evoked different responses on culpability and whether punishment should be used. A previous study by Weiner et al. used natural language processing to evaluate 500,000 notes from over 30,000 patients for language related to substance use based on NIDA’s “Words Matter” and found that out of patients with substance-use-related disorders, 62% had at least one note that contained stigmatizing language.

This study is the first to assess temporal trends in stigmatizing language over time in healthcare notes. Chart text search presents a novel method for healthcare systems to track efforts in reducing stigmatizing language in healthcare settings.

Methods: This study used Epic Systems SlicerDicer Software’s clinical note text search tool to conduct a retrospective analysis of healthcare workers’ notes receiving care in a large northeast US health system in Connecticut, Rhode Island, and New York from 1/1/2013 through 12/31/2025. The primary outcome was the percentage of notes with text mentions of stigmatizing versus non-stigmatizing terms. We estimated temporal trends using ordinary least squares linear regression, modeling calendar year as a continuous variable and the annual percentage of notes containing each term as the outcome measure. Percentages instead of net notes were used to account for an increase in notes over time, from 7,781,448 in 2013 to 22,978,653 in 2025, as the healthcare system visits increased.

Comparisons between comparable terms: “alcohol use disorder” versus “alcohol abuse,” “cocaine use disorder” versus “cocaine abuse,” and “polysubstance abuse” versus “polysubstance use” were used to account for temporal trends in substance use over the past 13 years. The mean terms and ratios between the terms (Appendix A) were also assessed to account for temporal changes in diagnoses of the use disorders.
Other non-preferred terms that do not have direct correlates were included in Appendix B (Febrez). Wildcard characters and alternate terms, including misspellings, were not included, other than opiate and opioid to account for the changing opioid supply.

Results: A total of 231,703,815 notes were analyzed using the clinical note text search tool. In the last 13 years, the chart language included both potentially stigmatizing and preferred correlate terms.

“Cocaine abuse” decreased from 0.117% of notes in 2013 to 0.058% of notes in 2025, while “cocaine use disorder” increased from < 0.001% of notes in 2013 to 0.127% of notes in 2025 (Figure 1a). Ordinary least squares (OLS) linear regression found a negative trend for “cocaine abuse” (β -0.0054, R2 0.919, SEslope +/-0.00048, pslope < 0.00001). OLS linear regression found a positive trend for “cocaine use disorder” (β +0.0097, R2 0.929, SEslope +/-0.00081, pslope < 0.00001).

The terms “opioid abuse” or “opioid abuse” decreased from 0.042% of notes in 2013 to 0.034% of notes in 2025, while “opioid use disorder” or “opiate use disorder” increased from 0.001% of notes in 2013 to 0.181% of notes in 2025 (Figure 1b). OLS linear regression found a negative trend for opiate or opioid abuse (β -0.00086, R2 0.469, SEslope +/-0.00028, pslope = 0.0097). OLS linear regression found a positive trend for opiate or opioid use disorder (β +0.0151, R2 0.993, SE slope +/-0.00039, pslope < 0.00001).

“Alcohol abuse” increased from 0.693% of notes in 2013 to 0.828% of notes in 2025, and “alcohol use disorder” increased from 0.003% of notes in 2013 to 0.457% of notes in 2025 (Figure 1c). OLS linear regression found a non-significant positive trend for “alcohol abuse” (β +0.0058, R2 0.040, SE slope +/-0.0085, pslope = 0.513). OLS linear regression found a positive trend for "alcohol use disorder" (β +.0383, R2 0.985, SE slope +/-0.00145, pslope < 0.00001).

“Polysubstance abuse” decreased from 0.22% of notes in 2013 to 0.114% of notes in 2025, while “polysubstance use” increased from 0.01% of notes in 2013 to 0.124% of notes in 2025 (Figure 1d). OLS linear regression found a negative trend for “polysubstance abuse” (β -0.0101, R2 0.878, SE slope +/-0.00113, pslope < 0.00001). OLS linear regression found a positive trend for "polysubstance use disorder "(β +.00969, R2 0.963, SE slope +/-0.00057, pslope < 0.00001).

For the year 2025, the ratio of cocaine use disorder to cocaine abuse terminology was 2.19, 5.32 for opioid use disorder, 0.55 for alcohol use disorder, and 1.09 for polysubstance use. All of these reflected an increase in the absolute value of the ratios since 2013.

Conclusion & Discussion: This is the largest charting analysis of healthcare notes assessing the language referring to patients with substance use disorders and the first study to incorporate temporal trends. Although there are still opportunities for improvement, the overall trend is that healthcare workers are moving toward documenting less stigmatizing language in charts for patients with substance use disorders. A limitation is that some of these charts may have copied forward terminology and diagnoses, which could mean underestimated new chart descriptions of substance use disorders using non-stigmatizing language. Other limitations include that there is no granular detail, including on the profession of the healthcare workers, which has been shown to affect results.

This study provides a framework for healthcare systems to track trends in chart language for referring to patients with substance use disorders and provides a framework for tracking quality improvement initiatives. Future directions include more granular tracking in different specialities, treatment settings (e.g., inpatient versus outpatient), and different healthcare professions for more targeted educational interventions. Reducing bias from healthcare providers towards patients with substance use disorders is a complex undertaking, and this study presents one way to track the effects of targeted efforts to reduce bias in the form of documentation.

References: 1. Yang LH, Wong LY, Grivel MM, Hasin DS. Stigma and substance use disorders: an international phenomenon. Curr Opin Psychiatry. 2017 Sep;30(5):378-388. doi: 10.1097/YCO.0000000000000351. PMID: 28700360; PMCID: PMC5854406.
2. Kelly JF, Westerhoff CM. Does it matter how we refer to individuals with substance-related conditions? A randomized study of two commonly used terms. Int J Drug Policy. 2010 May;21(3):202-7. doi: 10.1016/j.drugpo.2009.10.010. Epub 2009 Dec 14. PMID: 20005692.
3. Ashford RD, Brown AM, McDaniel J, Curtis B. Biased labels: An experimental study of language and stigma among individuals in recovery and health professionals. Subst Use Misuse. 2019;54(8):1376-1384. doi: 10.1080/10826084.2019.1581221. Epub 2019 Apr 4. PMID: 30945955; PMCID: PMC6510618.
4. Weiner SG, Lo YC, Carroll AD, Zhou L, Ngo A, Hathaway DB, Rodriguez CP, Wakeman SE. The Incidence and Disparities in Use of Stigmatizing Language in Clinical Notes for Patients With Substance Use Disorder. J Addict Med. 2023 Jul-Aug 01;17(4):424-430. doi: 10.1097/ADM.0000000000001145. Epub 2023 Jan 31. PMID: 37579100; PMCID: PMC10387497.
5. Bessette LG, Hauc SC, Danckers H, Atayde A, Saitz R. The associated press stylebook changes and the use of addiction-related stigmatizing terms in news media. Subst Abus. 2022;43(1):127-130. doi: 10.1080/08897077.2020.1748167. Epub 2020 Apr 29. PMID: 32348190.
McLaren N, Jones CM, Noonan R, Idaikkadar N, Sumner SA. Trends in stigmatizing language about addiction: A longitudinal analysis of multiple public communication channels. Drug Alcohol Depend. 2023 Apr 1;245:109807. doi: